东巴轮廓型字素可视化部件提取算法研究  

RESEARCH ON ALGORITHM FOR EXTRACTING VISUAL PART OF DONGBA CONTOUR GLYPHICS

作　　者：康厚良 杨玉婷 Kang Houliang;Yang Yuting(Sports Department,Suzhou Vocational University,Suzhou 215000,Jiangsu,China;School of Computer Engineering,Suzhou Vocational University,Suzhou 215000,Jiangsu,China)

机构地区：[1]苏州市职业大学体育部,江苏苏州215000 [2]苏州市职业大学计算机工程学院,江苏苏州215000

出　　处：《计算机应用与软件》2023年第2期197-205,共9页Computer Applications and Software

基　　金：苏州市职业大学引进人才科研启动金项目(201905000034);苏州市职业大学教改项目;云南省科学研究基金项目(2019J1152,2018JS748);国家社会科学基金项目(15BTY038)。

摘　　要：东巴文是一种十分原始的图画象形文字,提取东巴基本字素的可视化部件对于设计高效的东巴字识别算法,研究东巴字的结构、笔画、偏旁部首等有着非常重要的意义。因此,以部件表示法为理论基础,结合文字的可视化特征及东巴字特征曲线提取及简化算法给出了适用于东巴轮廓型字素的可视化部件提取算法,该算法简单、直观、易于实现,不仅解决了两类东巴基本字素在整体和局部上的一致性表示问题,还可以有效提高文字识别算法的健壮性和精确度。准确性、健壮性和对比性实验表明,该算法准确性高、健壮性好,具有良好的尺度、平移和旋转不变性。Dongba hieroglyph is a very primitive pictograph.Extracting visual parts is very important for designing efficient recognition algorithm,and studying the structure,strokes,and radicals of Dongba hieroglyphics.Therefore,we used part-based representations as the theoretical basis,combined with the visual feature of hieroglyph and the feature curve extraction and simplification algorithm,to give a visual part extraction algorithm for contour glyphics.The algorithm was intuitive,simple and easy to implement.It not only solved the problem of the overall and partial consistency of the two types of Dongba basic characters,but also effectively improved the robustness and accuracy of the recognition algorithm.The accuracy,robustness and comparative experiments show that the algorithm has high accuracy and robustness,and has good scale,translation and rotation invariance.

关 键 词：东巴文字 轮廓型字素 可视化特征 局部特征曲线 部件提取 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]